NEWS

xgxr 1.1.4

xgx_stat_ci: 1. for ordinal response, fix bug where response categories with count of zero were dropped and not displayed 2. for bins option, do a check to make sure the dataset can be divided into the specified number of bins using quantile. if not (say because one value is repeated many times), then throw an error that recommends the user set the breaks directly. 3. removed predict.nls, predictdf.nls, predictdf.polr from exports but add to registered functions in zzz.R to address issue with exports
xgx_scale_y_percentchangelog10 and xgx_scale_x_percentchangelog10: improved breaks, to use xgx_breaks_log10
xgx_breaks_log10: fixed error when range was c(0,0) so it now just plots an empty graph instead of throwing an error
xgx_scale_x_time_units: added minutes and seconds and require input to be the full time unit (e.g. year or years) instead of just a letter (y). Throws an error if not in the set (year, month, week, day, hour, minute, second)
xgx_breaks_time: added a few lines so preference is for grouping days by 30, 60, 90, instead of 70, 140, 210
xgx_scale_x_log10, xgx_scale_y_log10: no longer returns list. now returns ggproto
xgxr_overview vignette: removed a few functions from vignette no longer widely used - xgx_save, xgx_save_table, xgx_ xgx_summarize_data, xgx_summarize_covariates
xgx_annotate_status: added example for how to make it work with dates

xgx_conf_int: change df in xgx_conf_int from length(y) to length(y) - 1 for normal and lognormal distributions.
xgx_conf_int: Add warnings when length of y is 1, and change confidence interval from NaN to (-Inf, Inf) for normal and (0, Inf) for lognormal distributions.
xgx_conf_int: Add methods options for binomial and ordinal distributions. Change default method of ordinal to "goodman"
removed theoph.Rmd and sad_pkpd.Rmd from CRAN distribution (still available in GitHub)

change predictdf.polr so inputs match predictdf generic, to fix S3 method consistency warning

fix bug in xgx_stat_smooth / predictdf.nls that caused an error when formula had only one parameter to fit
add predict.nls function to produce predictions and confidence intervals for nls objects, similar to predict.lm for lm objects

xgx_stat_smooth() added to allow for nonlinear fitting + confidence intervals, including emax fitting, and ordinal response fitting
xgx_stat_ci() updated with binning by quantiles, and ordinal/multinomial confidence intervals
(BETA) xgx_auto_explore() function to automatically compile Rmd templates from xgx website!
xgx_save and xgx_dirs2char: Added error checking that dirs variable is list.
xgx_save: Enable change in font color and size for directories caption

changes to xgx_check_data() because some errors are new with regards to checking for negative values or NAs in the dataset

theoph.Rmd vignette, for data_pk, keep only rows where TIME>0. This fixed an error in the vignette that was introduced with the March 5, 2020 ggplot2 update

updating sad_pkpd vignette to show BLOQ data in red
minor change to xgx_check_data() text output
updating xgx_scale_x_reverselog() and xgx_scale_y_reverselog() to more easily control the labels and to allow user to set accuracy
xgx_breaks_time() - allow user to set target number of breaks xgx_scale_x_time_units()

fixed bug in xgx_theme_set() to make sure it calls xgx_theme() instead of theme_bw()
fixed bug in xgx_stat_ci() which changed default geom properties globally by mistake

fixed bug in xgx_geom_ci because it wasn't properly passing parameters to xgx_stat_ci
fixed bug in theoph.Rmd vignette to properly calculate the NOMTIME.
update theoph.Rmd and sad_pkpd.Rmd vignette to use TIME and NOMTIME both for confidence interval plots that overlay data

fixing to get in line with CRAN policies (e.g. example code cannot save files in the user's directory)